Producer for discriminator agains TTbar (as used in the EM channel)
Required config tags with an short example: { "MVATestMethodsInputQuantities": [ "0;pVecSum,lep1Pt:=pt_1,lep1MetMt:=mt_1,lep2Pt:=pt_2,lep2MetMt:=mt_2,pfMetPt:=met,mvaMetPt:=mvamet,pZetaMissVis,pZetaMiss,pZetaVis,nJets30:=njets,nBJets20:=nbtag,min_ll_jet_eta,lep1_centrality,lep2_centrality,delta_lep_centrality,lep1IsoOverPt:=iso_1", "1;pVecSum,lep1Pt:=pt_1,lep1MetMt:=mt_1,lep2Pt:=pt_2,lep2MetMt:=mt_2,pfMetPt:=met,mvaMetPt:=mvamet,pZetaMissVis,pZetaMiss,pZetaVis,nJets30:=njets,nBJets20:=nbtag,min_ll_jet_eta,lep1_centrality,lep2_centrality,delta_lep_centrality,lep1IsoOverPt:=iso_1" ], "MVATestMethodsMethods": [ "0;BDT", "0;BDT", "0;BDT", "1;BDT", "1;BDT" ], "MVATestMethodsNFolds": [ 3, 1, 1 ], "MVATestMethodsNames": [ "ggh_150_zXX", "ggh_250_zXX" ], "MVATestMethodsWeights": [ "/nfs/dust/cms/user/mschmitt/analysis/CMSSW_7_1_5/src/test_nfold/weights/T1_BDT_ggh_150_zXX.weights.xml", "/nfs/dust/cms/user/mschmitt/analysis/CMSSW_7_1_5/src/test_nfold/weights/T2_BDT_ggh_150_zXX.weights.xml", "/nfs/dust/cms/user/mschmitt/analysis/CMSSW_7_1_5/src/test_nfold/weights/T3_BDT_ggh_150_zXX.weights.xml", "/nfs/dust/cms/user/mschmitt/analysis/CMSSW_7_1_5/src/test_nfold/weights/T1_BDT_ggh_250_zXX.weights.xml", "/nfs/dust/cms/user/mschmitt/analysis/CMSSW_7_1_5/src/test_nfold/weights/T1_BDT_ggh_250_zXX.weights.xml" ] }
Detailed description: "MVATestMethodsInputQuantities": This is a string list containing training variables for different TMVA Factory instances. The number followed by ; defines the number of factory those variables belong to. "MVATestMethodsMethods": Number followed by ; defines the number ob factory those method belongs to "MVATestMethodsNFolds": This number indicates how much of the methods belong to one N-Fold Training: as in the example the 0;BDT methods belong to one single 3-Fold training and are treated accordingly "MVATestMethodsNames": This is a name base for n-fold and regular training. There will be exactly this name as variable and in case of N-Fold training several names derived from that one. For the 3-Fod training in the example: ggh_150_zXX -> {ggh_150_zXX, T1ggh_150_zXX, T2ggh_150_zXX, T3ggh_150_zXX} for a regular training there will not be a T1base_name present "MVATestMethodsWeights": Stringlist of weightfiles in the same order as the methods were specified
Inheritance diagram for MVATestMethodsProducer: